Enhancement of Detecting Permanent Water and Temporary Water in Flood Disasters by Fusing Sentinel-1 and Sentinel-2 Imagery Using Deep Learning Algorithms: Demonstration of Sen1Floods11 Benchmark Datasets

Abstract

Identifying permanent water and temporary water in flood disasters efficiently has mainly relied on change detection method from multi-temporal remote sensing imageries, but estimating the water type in flood disaster events from only post-flood remote sensing imageries still remains challenging. Research progress in recent years has demonstrated the excellent potential of multi-source data fusion and deep learning algorithms in improving flood detection, while this field has only been studied initially due to the lack of large-scale labelled remote sensing images of flood events. Here, we present new deep learning algorithms and a multi-source data fusion driven flood inundation mapping approach by leveraging a large-scale publicly available Sen1Flood11 dataset consisting of roughly 4831 labelled Sentinel-1 SAR and Sentinel-2 optical imagery gathered from flood events worldwide in recent years. Specifically, we proposed an automatic segmentation method for surface water, permanent water, and temporary water identification, and all tasks share the same convolutional neural network architecture. We utilize focal loss to deal with the class (water/non-water) imbalance problem. Thorough ablation experiments and analysis confirmed the effectiveness of various proposed designs. In comparison experiments, the method proposed in this paper is superior to other classical models. Our model achieves a mean Intersection over Union (mIoU) of 52.99%, Intersection over Union (IoU) of 52.30%, and Overall Accuracy (OA) of 92.81% on the Sen1Flood11 test set. On the Sen1Flood11 Bolivia test set, our model also achieves very high mIoU (47.88%), IoU (76.74%), and OA (95.59%) and shows good generalization ability.

Sentinel-1-Based Water and Flood Mapping: Benchmarking Convolutional Neural Networks Against an Operational Rule-Based Processing Chain

“…Sen1Floods11 is a dataset containing Sentinel-1 and Sentinel-2 data from various flood events, which has been recently used in a number of studies (e.g. [23], [24]). Details on the dataset can be found in Bonafilia et al [21].…”

“…Similar to Bai et at. [24], we choose DeepLabV3+ [44] (DL) as a second well-known model architecture to evaluate if a better segmentation can be achieved by employing different architectural concepts. DL applies atrous spatial pyramid pooling by using diluted convolution kernels with different rates to extract information on multiple scales simultaneously.…”

“…CNNs are a branch of artificial neural networks that have been shown to be capable of outperforming traditional image processing techniques in various fields ranging from image classification through object detection to image segmentation and have already been successfully applied to various problems in the earth observation domain, including flood and water mapping using either multi-spectral or SAR data (e.g. [19]- [24]).…”

“…Furthermore, they confirm that using VV-VH data leads to the best results for general surface water mapping. Last but not least, Bai et al [24] also used the dataset produced by Bonafilia et al [21] to test BASNet [36], an enhanced U-Net with a module to refine the residuals of the final feature maps. Furthermore, they assess various loss functions and investigate the impact of fusing Sentinel-1 and Sentinel-2 data, applying data augmentation and using various model architectures.…”

Automated Training Data Generation from Spectral Indexes for Mapping Surface Water Extent with Sentinel-2 Satellite Imagery at 10 m and 20 m Resolutions

“…The global availability of 10 m and 20 m Sentinel-2 imagery has enabled higher spatial and temporal resolution mapping especially within the challenging heterogeneous and urban landscapes [29][30][31]. Synthetic Aperture Radar is also used in conjunction with optical imagery to overcome issues with cloud cover and provide robust areal estimates [32][33][34][35][36][37].…”